Physical property data acquisition device and physical property data acquisition method

Abstract

A physical property data acquisition device disclosed in the present document comprises: a communication unit for receiving a plurality of pieces of numerical measurement data; and a control unit for generating preprocessed data by adding an index array to the plurality of pieces of numerical measurement data, inputting the preprocessed data to a regression model for deriving physical property data, and acquiring the physical property data as an output value on the basis of hyperparameters set in the regression model.

Description

Material property data acquisition device and material property data acquisition method

[0001] Cross-citation with related applications

[0002] This application claims the benefit of priority to Republic of Korea Patent Application No. 10-2023-0169806, filed November 29, 2023, the entire contents of which are incorporated herein by reference.

[0003] Technology field

[0004] The embodiments disclosed in this document relate to a material property data acquisition device and a material property data acquisition method.

[0005] Plastics are being used in a variety of fields, including automobiles and electronics, and product manufacturers must consider various material properties when designing products using plastics.

[0006] Accordingly, while traditionally evaluating material properties using various testing equipment, noise generated by the testing equipment and accumulated variation from repeated experiments have made it difficult to obtain high-quality material properties data. Therefore, obtaining high-quality material properties data using machine learning techniques is necessary.

[0007] According to one embodiment disclosed in this document, a property data acquisition device and a property data acquisition method are provided for inputting measurement data into a learning model to derive representative property information.

[0008] The technical problems of the embodiments disclosed in this document are not limited to the technical problems mentioned above, and other technical problems not mentioned will be clearly understood by those skilled in the art from the descriptions below.

[0009] A device for acquiring physical property data according to one embodiment includes a communication unit that receives a plurality of numerical measurement data, a control unit that adds an index column to the plurality of numerical measurement data to generate preprocessing data, inputs the preprocessing data into a regression model for deriving physical property data, and acquires the physical property data as an output value based on hyperparameters set in the regression model.

[0010] The above control unit can generate the preprocessing data by pairing the index column and the plurality of measurement data corresponding to the index column, for each of the plurality of measurement data.

[0011] The above control unit can generate a representative property curve based on the property data and transmit the representative property curve to an external device through the communication unit.

[0012] The above control unit can repeatedly derive the above physical property data and merge the above physical property data through data-based merging to derive the above representative physical property curve.

[0013] The above regression model can be constructed by combining a linear regression model and a Gaussian process regression model.

[0014] The control unit can acquire the physical property data by adjusting the parameters of the RBF (Radial Basis Function) kernel included in the Gaussian process regression model based on a user command for adjusting the hyper parameters received through the communication unit.

[0015] The above control unit can generate preprocessing data by adding an index column to Poisson ratio data included in the plurality of numerical measurement data, and obtain the physical property data based on the preprocessing data.

[0016] A method for obtaining material property data according to one embodiment includes receiving a plurality of numerical measurement data, generating preprocessing data by adding an index column to the plurality of numerical measurement data, inputting the preprocessing data into a regression model for deriving material property data, and obtaining the material property data as an output value based on hyperparameters set in the regression model.

[0017] Generating the above preprocessing data may include generating the preprocessing data by pairing the index column and the plurality of measurement data corresponding to the index column, for each of the plurality of measurement data.

[0018] A method for acquiring physical property data according to one embodiment may further include generating a representative physical property curve based on the physical property data and transmitting the representative physical property curve to an external device through the communication unit.

[0019] Generating the representative property curve may include repeatedly deriving the property data and merging the property data through data-based merging to derive the representative property curve.

[0020] The above regression model can be constructed by combining a linear regression model and a Gaussian process regression model.

[0021] A method for acquiring material property data according to one embodiment may further include acquiring the material property data by adjusting parameters of a Radial Basis Function (RBF) kernel included in the Gaussian process regression model based on a user command for adjusting the hyper parameter received through the communication unit.

[0022] Obtaining the above physical property data may include generating preprocessing data by adding an index column to Poisson ratio data included in the plurality of numerical measurement data, and obtaining the physical property data based on the preprocessing data.

[0023] According to a physical property data acquisition device according to one embodiment, high-quality physical property data can be acquired by filtering noise of physical property data through a machine learning technique.

[0024] Figure 1 illustrates a control block diagram of a physical property data acquisition device according to one embodiment.

[0025] FIG. 2 illustrates input and output values ​​of a machine learning model utilized in a physical property data acquisition device according to one embodiment.

[0026] FIG. 3 is a schematic flowchart illustrating a method for generating material data by a material data acquisition device according to one embodiment.

[0027] Figure 4 illustrates preprocessed data generated by a physical property data acquisition device according to one embodiment.

[0028] FIG. 5 illustrates measurement data utilized in a physical property data acquisition device according to one embodiment.

[0029] Figure 6 illustrates the result of matching measurement data with an index by a physical property data acquisition device according to one embodiment.

[0030] Figure 7 illustrates noise filtering performance according to hyper parameters in a physical property data acquisition device according to one embodiment.

[0031] Figures 8 to 10 illustrate representative property curves acquired by a property data acquisition device according to one embodiment.

[0032] Figure 11 illustrates a control flowchart of a method for acquiring physical property data according to one embodiment.

[0033] Hereinafter, various embodiments disclosed in this document will be described in detail with reference to the attached drawings. In this document, identical components in the drawings are designated by the same reference numerals, and redundant descriptions of identical components are omitted.

[0034] With respect to the various embodiments disclosed in this document, specific structural and functional descriptions are merely illustrative for the purpose of explaining the embodiments, and the various embodiments disclosed in this document may be implemented in various forms and should not be construed as being limited to the embodiments described in this document.

[0035] The expressions "first," "second," "first," or "second" used in various embodiments may describe various components, regardless of order and / or importance, and do not limit the components. For example, without departing from the scope of the embodiments disclosed herein, a first component may be renamed a second component, and similarly, a second component may also be renamed a first component.

[0036] The terms used in this document are intended solely to describe specific embodiments and may not be intended to limit the scope of other embodiments. Singular expressions may include plural expressions unless the context clearly dictates otherwise.

[0037] All terms used herein, including technical or scientific terms, have the same meaning as commonly understood by those of ordinary skill in the art of the embodiments disclosed herein. Terms defined in commonly used dictionaries may be interpreted as having the same or similar meaning in the context of the relevant technology, and unless explicitly defined herein, they shall not be interpreted in an idealized or overly formal sense. In some cases, even if a term is defined herein, it cannot be interpreted to exclude the embodiments disclosed herein.

[0038] The operating principle and embodiments of the present invention will be described with reference to the attached drawings below.

[0039] Figure 1 illustrates a control block diagram of a physical property data acquisition device according to one embodiment.

[0040] Referring to FIG. 1, a physical property data acquisition device (1) according to one embodiment includes a control unit (100) including at least one processor (110) and a memory (120) and a communication unit (200), and can transmit physical property data to the outside by communicating with an external device (2) through the communication unit (200).

[0041] A physical property data acquisition device (1) according to one embodiment may be implemented as a server that acquires physical property data based on machine learning, and specifically, may be implemented as various computing devices such as a workstation, a cloud, a data drive, a data station, etc. In addition, the physical property data acquisition device (1) may be implemented as one or more server devices that are physically or logically separated based on function, detailed configuration of function, or data, and may transmit and receive data and process the transmitted and received data through communication between the respective server devices.

[0042] An external device (2) communicating with a material property data acquisition device (1) may include a user terminal and an external server device that receive material property data generated by the material property data acquisition device (1).

[0043] Specifically, when the external device (2) is a user terminal, the control unit (100) of the physical property data acquisition device (1) can transmit the physical property data generation result to the user terminal so that the user can check it. At this time, the user terminal may include, but is not limited to, a personal computer, a terminal, a portable telephone, a smart phone, a handheld device, a wearable device, etc.

[0044] In addition, when the external device (2) is an external server device, the external server device can be implemented as various computing devices, and the external server device can be implemented as one or more external server devices, and data can be transmitted and received and the transmitted and received data can be processed through communication between each external server device.

[0045] A physical property data acquisition device (1) according to one embodiment may mean any electronic device including a processor (110) and a memory (120). Each component of the physical property data acquisition device (1) will be described in detail below.

[0046] The communication unit (200) may include a wireless communication unit (210) and a wired communication unit (220) to communicate with an external device (2). The communication unit (200) may receive numerical measurement data to be used as learning data from a separately provided external device (2), or may transmit and receive programs and various data for machine learning.

[0047] The wireless communication unit (210) may include at least one of a short-range communication module and a long-range communication module.

[0048] The short-range communication module can communicate with an external device (2) adjacent to the physical property data acquisition device (1) using a short-range communication method. Here, the short-range communication module can utilize one of the following communication methods: Bluetooth, Bluetooth low energy, infrared data association (IrDA), Zigbee, Wi-Fi, Wi-Fi direct, Ultra Wideband (UWB), or near field communication (NFC).

[0049] The remote communication module may include a communication module that performs various types of remote communication and may include a mobile communication unit. The mobile communication unit may transmit and receive a wireless signal with at least one of a base station, an external terminal, and an external device (2) on a mobile communication network. In addition, the remote communication module may communicate with an external device (2) or an external device (2) such as another electronic device through a nearby access point (AP). The access point (AP) may connect a local area network (LAN) to which the physical data acquisition device (1) is connected to a wide area network (WAN) to which a communication server is connected. Accordingly, the physical data acquisition device (1) may be connected to the communication server through the wide area network (WAN) and communicate with the external device (2).

[0050] The wired communication unit (220) can connect to a wired communication network and communicate with an external device (2) through the wired communication network. For example, the wired communication unit (220) can connect to a wired communication network through Ethernet (IEEE 802.3 technology standard) or connect to a wired communication network through CAN communication, and transmit and receive data with the external devices (2) through the wired communication network.

[0051] A physical property data acquisition device (1) according to one embodiment may include an input / output interface (not shown). An interface may be provided that connects an input device (not shown) such as a keyboard, mouse, or touch panel, an output device (not shown) such as a display (not shown), and a processor (110) to transmit and receive data.

[0052] The memory (120) can store various types of information required for driving the physical property data acquisition device (1). Specifically, the memory (120) can store an operating system and a program required for driving the physical property data acquisition device (1), or store data required for driving the physical property data acquisition device (1).

[0053] Specifically, the memory (120) can store various programs related to measurement data preprocessing, representative material curve derivation algorithms, and machine learning algorithms. In addition, the memory (120) can store feature data utilized in machine learning.

[0054] The memory (120) may include volatile memory (120) such as Static Random Access Memory (S-RAM) and Dynamic Random Access Memory (D-RAM) for temporarily storing data. In addition, the memory (120) may include nonvolatile memory (120) such as Read Only Memory (ROM), Erasable Programmable Read Only Memory (EPROM), and Electrically Erasable Programmable Read Only Memory (EEPROM) for long-term storage of data.

[0055] The processor (110) outputs a control signal to control the overall physical data acquisition device (1). The processor (110) may include one or more central processing units (CPUs) and graphics processing units (GPUs). In this case, the processor (110) may be implemented as an array of a plurality of logic gates, or may be implemented as a combination of a general-purpose microprocessor (110) and a memory (120) storing a program that can be executed on the microprocessor (110).

[0056] The aforementioned memory (120) and processor (110) can be included in the control unit (100), and the control unit (100) can control the aforementioned components to obtain high-quality physical property data.

[0057] Specifically, the control unit (100) can generate preprocessed data by adding an index column to a plurality of numerical measurement data. That is, the control unit (100) can preprocess the measurement data so that the data can be effectively learned and filtered in a machine learning model, and can separate the measurement data representing a plurality of characteristics by adding an index column.

[0058] That is, the control unit (100) can generate preprocessing data by pairing the index column and the plurality of measurement data corresponding to the index column, one-to-one, for each of the plurality of measurement data.

[0059] Thereafter, the control unit (100) can input preprocessed data into the regression model and obtain material property data as output values ​​based on the hyperparameters set in the regression model. At this time, the hyperparameters refer to variables that are adjusted by the learning algorithm of the model itself in the machine learning model, and the hyperparameters are set before the model is trained and can be used to control and adjust the model's learning process.

[0060] The control unit (100) can generate a representative property curve based on the property data and transmit the representative property curve to an external device (2) via the communication unit (200). Here, the representative property curve means a curve representing a specific material or property, and may include a tensile stress-strain curve and a torque-z displacement curve, etc.

[0061] In addition, the control unit (100) can repeatedly derive material property data and merge the material property data through data-based merging to derive a representative material property curve, thereby obtaining material property data with high versatility and consistency.

[0062] In addition, the control unit (100) can acquire material property data by adjusting the parameters of the RBF (Radial Basis Function) kernel included in the Gaussian process regression model based on a user command for adjusting hyper parameters received through the communication unit (200).

[0063] That is, users can adjust the hyperparameters of the machine learning model, thereby controlling the noise filtering and data merging performance.

[0064] Accordingly, the physical property data acquisition device (1) can acquire high-quality physical property data by utilizing a machine learning model, thereby reducing the time and cost required in the process of acquiring physical property data.

[0065] Fig. 2 illustrates input and output values ​​of a machine learning model utilized in a physical property data acquisition device (1) according to one embodiment.

[0066] The control unit (100) can train a machine learning model using the material property measurement data included in a plurality of numerical measurement images and the signal data of the testing machine as training data (a), and can derive representative material property data as output data (b).

[0067] When the control unit (100) includes an artificial intelligence processor (110) (e.g., NPU) for training a machine learning model, the processor (110) can train an artificial neural network by utilizing weight data stored in the memory (120) as training data for the machine learning model.

[0068] Examples of learning algorithms include, but are not limited to, supervised learning, unsupervised learning, semi-supervised learning, or reinforcement learning.

[0069] An artificial neural network included in a machine learning model may be composed of multiple neural network layers. Each of the multiple neural network layers has multiple weight values ​​and performs neural network operations through operations between the calculation results of the previous layer and the multiple weights. The multiple weights of the multiple neural network layers can be optimized based on the learning results of an artificial intelligence model. For example, during the learning process, the multiple weights may be updated so that the loss or cost values ​​obtained from the artificial intelligence model are reduced or minimized.

[0070] The artificial neural network may include a deep neural network (DNN), for example, a convolutional neural network (CNN), a deep neural network (DNN), a recurrent neural network (RNN), a restricted boltzmann machine (RBM), a deep belief network (DBN), a bidirectional recurrent deep neural network (BRDNN), or deep Q-networks, but is not limited to the examples described above.

[0071] The control unit (100) can learn the relationship between the material property measurement data and testing machine signal data included in the numerical measurement data and the representative material property data based on the selected artificial intelligence model.

[0072] Figure 3 illustrates a schematic flowchart of a physical property data acquisition device (1) according to one embodiment generating physical property data.

[0073] The configurations 101 to 105 in FIG. 3 are implemented in the form of software blocks, and can be stored in memory (120) and executed by the processor (110).

[0074] First, referring to FIG. 3, the control unit (100) can receive a plurality of numerical measurement data through the communication unit (200) and store the received measurement data in the measurement data DB (121). At this time, the plurality of numerical measurement data can include measurement values ​​used to measure the physical properties of a material, and can include, for example, tensile strength, tensile strain, hardness, toughness, thermal conductivity, and coefficient of thermal expansion.

[0075] The preprocessing data generation unit (101) of the control unit (100) may perform data preprocessing to process received measurement data into learning data. Data preprocessing refers to the process of converting raw measurement data into a form applicable to a machine learning model, and may include all processes for cleaning the data and enabling the model to easily learn patterns.

[0076] Specifically, the control unit (100) can preprocess the data by transforming the numerical measurement data, replacing the x value with an index, and making the y value correspond one-to-one with the existing data. In addition, the control unit (100) can preprocess the data by removing missing values ​​and strings from the numerical measurement data.

[0077] Thereafter, the machine learning model learning unit (102) of the control unit (100) can learn the model based on preset hyperparameters and selected features. Specifically, the control unit (100) can learn the parameters of the model by using the selected features as input data to filter measurement data and generate physical property data.

[0078] A material property data acquisition device (1) according to one embodiment can utilize an algorithm that combines a linear regression model and a Gaussian process regression model to acquire high-quality material property data.

[0079] Specifically, the control unit (100) can improve the performance of the model by combining the flexibility of the Gaussian process with the simplicity and interpretability of the linear regression model by combining a linear regression model and a Gaussian process regression model.

[0080] Here, a linear regression model models the linear relationship between a dependent variable and one or more independent variables, and can refer to an algorithm that finds a single straight line that best describes the data. Specifically, a linear regression model is used to predict continuous values ​​for given input variables, and model parameters can typically be estimated using methods such as the least squares method.

[0081] Furthermore, the Gaussian process regression model is a probabilistic model that models the relationships between given data points using a Gaussian process. It can model data uncertainty by predicting the probability distribution for given inputs. The Gaussian process regression model can estimate the relationships between data points using a kernel function that depends on the training data.

[0082] Afterwards, the control unit (100) can store the learned machine learning model in the machine learning learning model DB (122).

[0083] The property data generation unit (103) of the control unit (100) can generate property data corresponding to the output value of the learning model. Specifically, the control unit (100) can utilize the smooth curve derivation characteristic of the RBF (Radial Basis Function) kernel of the Gaussian process regression model, and can generate property data by performing noise filtering and data merging regardless of the form of the numerical measurement data, which is the input value.

[0084] Thereafter, the representative property curve generation unit (104) of the control unit (100) can generate a representative property curve by merging repeated experimental data. Specifically, the control unit (100) performs data-based merging rather than formula-based merging, thereby facilitating application and alignment in other fields.

[0085] The control unit (100) can generate a representative material property curve and transmit it to an external device (2), so that the material properties of the corresponding material can be efficiently provided to a user who wishes to utilize the material property data.

[0086] As described above, the physical property data acquisition device (1) according to one embodiment applies machine learning to preprocess numerical measurement data and generates physical property data based on preset hyperparameters, thereby enabling the acquisition of highly reliable physical property data. This makes physical property data acquisition easier than before, and ensures the quality of physical property data that was previously unattainable.

[0087] Fig. 4 illustrates preprocessing data generated by a physical property data acquisition device (1) according to one embodiment.

[0088] Referring to FIG. 4, the control unit (100) can receive strain-standard force data as measurement data, and the strain-standard force data means a set of data points for strain and applied force obtained from a tensile test of a material.

[0089] Fig. 4(a) refers to raw measurement data received through the communication unit (200), and Fig. 4(b) refers to preprocessed data preprocessed by the control unit (100).

[0090] Typically, strain-force data is expressed in pairs and visualized in graphs to illustrate material properties. Accordingly, the raw measurement data (a) received via the communication unit (200) may be composed of strain data and applied force data in pairs.

[0091] However, a physical property data acquisition device (1) according to one embodiment can acquire preprocessed data (b) by matching each data characteristic included in raw measurement data (a) to a unified index for effective filtering and machine learning.

[0092] For example, in Fig. 4(a), the strain values ​​are formed in the order of 0.0005 -0.0004, 0.0014, and 0.0009 from the top, and can be received by matching them to 9.00525, 9.29375, 12.87025, and 15.697 in the applied force (Standard force), respectively.

[0093] The control unit (100) can add an index column (c) to each characteristic value, and in the case of the strain value, 0.0005 can be matched to index 1 in order from the top, -0.0004 can be matched to index 2, 0.0014 can be matched to index 3, and 0.0009 can be matched to index 4. In addition, in the case of the applied force (Standard force) value, the control unit (100) can be matched to index 1 in order from the top, 9.00525 can be matched to index 1, 9.29375 can be matched to index 2, 12.87025 can be matched to index 3, and 15.697 can be matched to index 4.

[0094] Accordingly, the performance of data analysis and modeling can be improved because information about measurement data points can be easily derived based on the index.

[0095] FIG. 5 illustrates measurement data utilized in a physical property data acquisition device (1) according to one embodiment, and FIG. 6 illustrates a result of matching measurement data with an index by a physical property data acquisition device (1) according to one embodiment.

[0096] First, (a) of Fig. 5 shows the relationship between the force applied to the material (standard force) and the elongation of deformation (deformation elongation), and may mean a graph that visually represents data obtained from a tensile test of the material.

[0097] Figure 5 (a) represents a linear region where the initial increase is linear, and after a certain point, the deformation may remain constant or slightly decrease. However, if (a) is input directly into a learning model without any preprocessing, noise may affect the correlation between each data and the output value.

[0098] With continued reference to FIG. 6, a physical property data generation device according to one embodiment can match each characteristic value to a unified index and perform learning by minimizing noise.

[0099] That is, the control unit (100) can match data regarding deformation elongation characteristics to indices 0 to 600, as shown in (a-1) of FIG. 6, and thereby learn the deformation elongation characteristics.

[0100] Likewise, the control unit (100) can match data on the force (standard force) characteristics applied to the material to indices 0 to 600, as shown in (a-2) of FIG. 6, and thereby learn the force characteristics applied to the material.

[0101] Thereafter, the control unit (100) can learn each characteristic matched to the index using a machine learning algorithm, merge data regarding each characteristic, and derive a representative material property curve based on the merged data.

[0102] Fig. 7 illustrates noise filtering performance according to hyper parameters in a physical property data acquisition device (1) according to one embodiment.

[0103] A physical property data acquisition device (1) according to one embodiment can adjust the noise filtering degree of physical property data based on hyperparameters input by the user. Accordingly, since the user can intuitively set hyperparameters based on indices, noise filtering and data merging performance can be improved overall.

[0104] Specifically, the graph in Fig. 7 shows the noise filtering performance according to hyper parameters when learning is performed on high speed tensile test data, and the graph indicates the size of the force (standard force) applied to the material according to the index.

[0105] Figure 7 (a) may refer to a case where a specific hyperparameter is set to 10 for high-speed tensile test data, and Figure 7 (b) may refer to a case where the same hyperparameter as in (a) is set to 1000. In this case, the numerical values ​​of the hyperparameters are exemplary and may change depending on the learning model and learning data.

[0106] Referring to (a), the center line may be displayed thicker due to noise compared to (b), and in (b), the center line may be displayed thinner due to reduced noise compared to (a).

[0107] In this way, the physical property data acquisition device (1) according to one embodiment can receive hyper parameters from a user and adjust the performance of noise filtering based on the hyper parameters, thereby having the effect of deriving optimal physical property data.

[0108] Figures 8 to 10 illustrate representative property curves acquired by a property data acquisition device (1) according to one embodiment.

[0109] The control unit (100) can generate a representative property curve for each material based on property data derived through machine learning. Here, the representative property curve may refer to a representative graph representing the physical, chemical, and mechanical properties of a specific material.

[0110] Specifically, Fig. 8 represents representative material property curves derived in different ways based on data of stress (load) and deformation elongation (deformation elongation) obtained from a tensile test.

[0111] Since (a) of Fig. 8 corresponds to raw measurement data, it may contain a lot of noise and obtain an uneven representative material property curve.

[0112] In addition, (b) of FIG. 8 may represent a graph obtained by obtaining a representative property curve using a formula-based data merging method, and (c) of FIG. 8 may represent a graph obtained by obtaining a representative property curve using a data-based data merging method by a control unit (100).

[0113] Fig. 8 (b) and Fig. 8 (c) may show similar trends because they obtained physical property data using a machine learning technique, but since the graph (c) merges data based on the data, it has a superior general-purpose effect compared to the graph (b).

[0114] Next, Fig. 9 represents a representative material property curve derived by the control unit (100) based on the torque-z displacement data obtained from the tensile test.

[0115] Since (a) of Fig. 9 corresponds to raw measurement data, it contains a lot of noise and an uneven representative property curve can be obtained.

[0116] On the other hand, (b) of Fig. 9 may mean a graph obtained by obtaining a representative property curve by a data-based data merging method by the control unit (100).

[0117] Comparing (a) and (b) of Fig. 9, (b) of Fig. 9 has obtained material property data using a machine learning technique, so it has many linear regions compared to the raw measurement data, making it easy to identify trends, and it has the effect of making it easy to understand and predict the characteristics of the material through trends.

[0118] Next, Fig. 10 represents a representative material property curve for Poisson's ratio derived based on the X-strain-Y-strain data obtained from the tensile test.

[0119] Since (a) of Fig. 10 corresponds to raw measurement data, it contains a lot of noise and an uneven representative property curve can be obtained.

[0120] On the other hand, (b) of Fig. 10 may mean a graph obtained by obtaining a representative property curve by a data-based data merging method by the control unit (100).

[0121] Comparing (a) and (b) of Fig. 10, (b) of Fig. 10 is obtained by using a machine learning technique to obtain material property data, so it can be expressed in a linear region compared to raw measurement data, and the characteristics of the material can be easily identified.

[0122] Ultimately, the physical property data acquisition device (1) according to one embodiment acquires preprocessed data by matching a representative physical property curve to an index, and merges the results of inputting the preprocessed data into a machine learning model based on data, thereby acquiring physical property data with less influence of noise and high versatility.

[0123] Figure 11 illustrates a control flowchart of a method for acquiring physical property data according to one embodiment.

[0124] Referring to FIG. 11, the control unit (100) can receive a plurality of numerical measurement data via the communication unit (200) (1100). Thereafter, the control unit (100) can add an index column to the received measurement data (1110), and specifically, can generate preprocessing data by pairing the corresponding measurement data with the index column (1120). In other words, the control unit (100) can separate the measurement data expressed as data pairs and generate preprocessing data as index-data pairs.

[0125] Thereafter, the control unit (100) can receive hyperparameters for deriving physical property data (1130). At this time, the hyperparameters may refer to values ​​preset by the user, and machine learning model training can be performed based on the hyperparameters.

[0126] The control unit (100) can input preprocessing data into a regression model for deriving material property data (1140), and the control unit (100) can determine whether the material property data derived based on hyperparameters is above a preset quality standard (1150). Specifically, a reference value for whether the material property data sufficiently reflects the characteristics of the material may be stored in the memory (120), and the control unit (100) can compare the material property data with the reference value to determine whether it is above the quality standard.

[0127] If the control unit (100) determines that the physical property data derived based on the hyperparameters falls below the preset quality standard (No in 1150), it may generate a notification requesting modification of the hyperparameters (1160). However, the notification requesting modification of the hyperparameters may be omitted and may be directly modified by the user.

[0128] Thereafter, if the control unit (100) determines that the material property data derived based on the hyper parameters is above a preset quality standard (example of 1150), it can generate a representative material property curve based on the material property data (1170) and transmit the representative material property curve to an external device (2) through the communication unit (200).

[0129] In this way, according to the physical property data acquisition device (1) according to one embodiment, a machine learning-based regression technique can be utilized to acquire physical property data, so there is an effect of enabling the generation of more trend-oriented physical property data.

[0130] Meanwhile, the disclosed embodiments may be implemented in the form of a recording medium storing computer-executable instructions. The instructions may be stored in the form of program code, and when executed by a processor, may generate program modules to perform the operations of the disclosed embodiments. The recording medium may be implemented as a computer-readable recording medium.

[0131] Computer-readable storage media include all types of storage media that store instructions that can be deciphered by a computer. Examples include read-only memory (ROM), random access memory (RAM), magnetic tape, magnetic disks, flash memory, and optical data storage devices.

[0132] Additionally, a computer-readable recording medium may be provided in the form of a non-transitory storage medium. Here, the term "non-transitory storage medium" simply means a tangible device that does not contain signals (e.g., electromagnetic waves). This term does not distinguish between cases where data is permanently stored in the storage medium and cases where data is temporarily stored. For example, a "non-transitory storage medium" may include a buffer in which data is temporarily stored.

[0133] According to one embodiment, the method according to the various embodiments disclosed in the present document may be provided as included in a computer program product. The computer program product may be traded as a product between a seller and a buyer. The computer program product may be distributed in the form of a machine-readable recording medium (e.g., compact disc read only memory (CD-ROM)), or may be distributed online (e.g., downloaded or uploaded) via an application store (e.g., Play Store™) or directly between two user devices (e.g., smartphones). In the case of online distribution, at least a portion of the computer program product (e.g., a downloadable app) may be at least temporarily stored or temporarily generated in a machine-readable recording medium, such as a memory of a manufacturer's server, an application store's server, or an intermediary server.

[0134] The above has illustrated and described specific embodiments. However, the invention is not limited to the above-described embodiments, and those skilled in the art will appreciate that various modifications and implementations can be made without departing from the spirit and scope of the invention as set forth in the claims below.

[0135] [Explanation of symbols]

[0136] 1: Physical property data acquisition device

[0137] 2: External devices

[0138] 100: Control Unit

[0139] 110: Processor

[0140] 120: Memory

[0141] 121: Measurement Data DB

[0142] 122: Machine Learning Model DB

[0143] 200: Communications Department

[0144] 210: Wireless Communications Department

[0145] 220: Wired Communications Department

Claims

1. A communication unit for receiving multiple numerical measurement data; and A property data acquisition device including a control unit that generates preprocessing data by adding an index column to the plurality of numerical measurement data, inputs the preprocessing data into a regression model for deriving property data, and acquires the property data as an output value based on hyper parameters set in the regression model.

2. In claim 1, The above control unit, A property data acquisition device that generates the preprocessing data by pairing the index column and the plurality of measurement data corresponding to the index column, for each of the plurality of measurement data.

3. In claim 1, The above control unit, A property data acquisition device that generates a representative property curve based on the above property data and transmits the representative property curve to an external device through the communication unit.

4. In claim 3, The above control unit, A property data acquisition device that repeatedly derives the above property data and merges the property data through data-based merging to derive the representative property curve.

5. In claim 1, The above regression model is, A material data acquisition device configured by combining a linear regression model and a Gaussian process regression model.

6. In claim 5, The above control unit, A material data acquisition device that acquires the material data by adjusting the parameters of the RBF (Radial Basis Function) kernel included in the Gaussian process regression model based on a user command for adjusting the hyper parameters received through the communication unit.

7. In claim 1, The above control unit, A property data acquisition device that generates preprocessing data by adding an index column to Poisson ratio data included in the above plurality of numerical measurement data, and acquires the property data based on the preprocessing data.

8. Receive multiple numerical measurement data; Generate preprocessing data by adding an index column to the above multiple numerical measurement data; The above preprocessing data is input into a regression model for deriving physical property data; A method for obtaining material data, comprising: obtaining the material data as an output value based on the hyperparameters set in the regression model.

9. In claim 8, Generating the above preprocessing data is as follows: A method for obtaining physical property data, comprising: generating the preprocessing data by pairing the index column and the plurality of measurement data corresponding to the index column, for each of the plurality of measurement data.

10. In claim 8, A method for obtaining material data, further comprising: generating a representative material curve based on the material data and transmitting the representative material curve to an external device through the communication unit.

11. In claim 10, Generating the above representative property curves is: A method for obtaining material data, comprising: repeatedly deriving the material data above, and merging the material data through data-based merging to derive the representative material curve.

12. In claim 8, The above regression model is, A method for obtaining material property data by combining a linear regression model and a Gaussian process regression model.

13. In claim 12, A method for obtaining material data, further comprising: obtaining the material data by adjusting parameters of an RBF (Radial Basis Function) kernel included in the Gaussian process regression model based on a user command for adjusting the hyper parameters received through the communication unit.

14. In claim 8, Obtaining the above physical property data is as follows: A method for obtaining property data, comprising: generating preprocessing data by adding an index column to Poisson ratio data included in the plurality of numerical measurement data, and obtaining the property data based on the preprocessing data.

Images